Exploring investment requirements for energy efficiency upgrades in the private rental sector

Kren, Janez; Kenny, Eoin; O'Toole, Conor; Shiel, Eva; Slaymaker, Rachel Research Report Exploring investment requirements for energy efficiency upgrades in the private rental sector Research Series, No. 205 Provided in Cooperation with: The Economic and Social Research Institute (ESRI), Dublin Suggested Citation: Kren, Janez; Kenny, Eoin; O'Toole, Conor; Shiel, Eva; Slaymaker, Rachel (2025) : Exploring investment requirements for energy efficiency upgrades in the private rental sector, Research Series, No. 205, The Economic and Social Research Institute (ESRI), Dublin, https://doi.org/10.26504/RS205 This Version is available at: https://hdl.handle.net/10419/322443 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Exploring investment requirements for energy efficiency upgrades in the private rental sector JANEZ KREN, EOIN KENNY, CONOR O’TOOLE, EVA SHIEL AND RACHEL SLAYMAKER ESRI RESEARCH SERIES Number 205, February 2025 EXPLORING INVESTMENT REQUIREMENTS FOR ENERGY EFFICIENCY UPGRADES IN THE PRIVATE RENTAL SECTOR Janez Kren Eoin Kenny Conor O’Toole Eva Shiel Rachel Slaymaker February 2025 ESRI RESEARCH SERIES NUMBER 205 Available to download from ww.esri.ie https://doi.org/10.26504/RS205 © The Economic and Social Research Institute Whitaker Square, Sir John Rogerson’s Quay, Dublin 2 This Open Access work is licensed under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium, provided the original work is properly credited. ABOUT THE ESRI The Economic and Social Research Institute (ESRI) advances evidence-based policymaking that supports economic sustainability and social progress in Ireland. ESRI researchers apply the highest standards of academic excellence to challenges facing policymakers, focusing on ten areas of critical importance to 21st century Ireland. The Institute was founded in 1960 by a group of senior civil servants led by Dr T.K. Whitaker, who identified the need for independent and in-depth research analysis. Since then, the Institute has remained committed to independent research and its work is free of any expressed ideology or political position. The Institute publishes all research reaching the appropriate academic standard, irrespective of its findings or who funds the research. The ESRI is a company limited by guarantee, answerable to its members and governed by a Council, comprising up to 14 representatives drawn from a crosssection of ESRI members from academia, civil services, state agencies, businesses and civil society. Funding for the ESRI comes from research programmes supported by government departments and agencies, public bodies, competitive research programmes, membership fees and an annual grant-in-aid from the Department of Public Expenditure NDP Delivery and Reform. Further information is available at www.esri.ie. THE AUTHORS Conor O’Toole is an Associate Research Professor at the ESRI and an Adjunct Professor at Trinity College Dublin (TCD). Janez Kren and Rachel Slaymaker are Research Officers at the ESRI and hold Adjunct Assistant Professor positions at TCD. Eoin Kenny and Eva Shiel were previously Research Assistants at the ESRI. This report has been accepted for publication by the Institute, which does not itself take institutional policy positions. The report has been peer reviewed prior to publication. The authors are solely responsible for the content and the views expressed. Table of contents| i TABLE OF CONTENTS LIST OF ABBREVIATIONS .......................................................................................................................... v EXECUTIVE SUMMARY ............................................................................................................................ vi CHAPTER 1 INTRODUCTION ..................................................................................................................... 1 CHAPTER 2 BACKGROUND AND CONTEXT ............................................................................................... 5 2.1 The Irish housing market ..................................................................................................... 5 2.2 International literature ........................................................................................................ 9 2.3 Summary of relevant Irish research .................................................................................. 12 CHAPTER 3 PROFILING THE ENERGY EFFICIENCY OF THE PRIVATE RENTAL SECTOR ............................ 14 3.1 Introduction ....................................................................................................................... 14 3.2 RTB registrations data ....................................................................................................... 14 3.3 SEAI BER database ............................................................................................................. 19 3.4 Census reconciliation and comparative analysis ............................................................... 25 CHAPTER 4 UNDERSTANDING INVESTMENT EXPENDITURE NEEDS ...................................................... 32 4.1 Introduction ....................................................................................................................... 32 4.2 Cost of efficiency upgrade data ......................................................................................... 32 4.3 Estimation of upgrade cost per dwelling ........................................................................... 37 4.4 Towards an aggregate cost of PRS dwelling upgrades ...................................................... 41 CHAPTER 5 INVESTMENT BARRIERS AND THE LANDLORD STRUCTURE ................................................ 46 5.1 Data and demographic profile of household landlords .................................................... 47 5.2 Simulating hypothetical financing gaps ............................................................................. 57 CHAPTER 6 CONCLUDING REMARKS...................................................................................................... 68 6.1 Findings on investment needs ........................................................................................... 68 6.2 Findings on the landlord structure and financial capacity to invest ................................. 69 REFERENCES .............................................................................................................................. 72 APPENDIX A ADDITIONAL RESULTS ........................................................................................................ 75 APPENDIX B PROTECTED HERITAGE BUILDINGS .................................................................................... 78 APPENDIX C REGRESSION RESULTS FOR MISSING RTB BER RATINGS ................................................... 80 ii|Investment requirements for energy efficiency upgrades in the rental sector LIST OF TABLES Table 3.1 RTB-registered properties across registration type ............................................. 15 Table 3.2 Characteristics of properties with BER compared to without BER ....................... 26 Table 3.3 Characteristics of properties with BER compared to without BER ....................... 26 Table 4.1 Number of observations by dataset ..................................................................... 34 Table 4.2 Example using RTB data and quadratic fit upgrade costs .................................... 43 Table 4.3 Estimated aggregate costs of upgrade to B* for all G–C1 dwellings (€mn) ......... 44 Table 5.1 Summary of landlord annual household income data – survey year 2020 .......... 52 Table 5.2 Summary of landlord annual rental income data – survey years 2018/2020 ...... 54 Table 5.3 Definitions of wealth variables ............................................................................. 55 Table 5.4 Summary of main wealth variables ...................................................................... 55 Table 5.5 Summary of savings (€) ........................................................................................ 56 Table 5.6 Summary of indebtedness .................................................................................... 56 Table 5.7 Proportion of landlords unable to finance the investment activity by financial measure and investment size ............................................................................... 58 Table 5.8 Median investment gap of landlords unable to finance the investment activity by financial measure and investment size ................................................................ 59 Table 5.9 Median monthly repayment amount of landlords who require a loan to cover the investment gap ..................................................................................................... 60 Table 5.10 DSR for landlords with existing debt and repayments as per investment gaps ... 62 Table 5.11 Loan-to-value ratio (portfolio level) for landlords with existing debt and borrowing above loan amounts ............................................................................................. 62 Table A.1 Concordance of dwelling types ............................................................................ 75 Table A.2 Mean estimates of the rental housing stock size ................................................. 75 Table A.3 Number of observations in upgrade cost data, preand post-upgrade ............... 76 Table A.4 Cost of upgrade estimation results ...................................................................... 77 Table A.5 Estimates of aggregate costs of upgrade to b* for all dwellings with ber g to c1, without accounting for heritage buildings ........................................................... 77 Table B.1 Estimated percentage of heritage buildings by county and dwelling type .......... 79 Table C.1 Characteristics of properties with BER compared to without BRT ....................... 80 Investment requirements for energy efficiency upgrades in the rental sector|iii LIST OF FIGURES Figure 2.1 Tenure structure of Irish households from Census data ........................................ 6 Figure 2.2 BER structures for different groups of rental sector (% of total) ........................... 8 Figure 2.3 Property age by landlord type (% of total) ............................................................. 8 Figure 3.1 RTB-registered properties across dwelling type................................................... 16 Figure 3.2 RTB-registered properties by county ................................................................... 16 Figure 3. 3 RTB-registered properties by number of bedrooms ............................................ 17 Figure 3. 4 RTB-registered properties with self-reported BER ............................................... 17 Figure 3. 5 Distribution of self-reported BER of RTB-registered properties .......................... 18 Figure 3.6 BER distribution by county and dwelling type ...................................................... 19 Figure 3. 7 Number of assessments by assessment purpose ................................................. 20 Figure 3.8 Number of assessments by year of assessment ................................................... 21 Figure 3. 9 Number of assessments by dwelling type ............................................................ 22 Figure 3.10 Number of assessments by year of construction ................................................. 22 Figure 3.11 Number of assessments by floor area .................................................................. 23 Figure 3.12 BER ratings distribution of assessments ............................................................... 23 Figure 3.13 BER assessments distribution by year of assessment .......................................... 24 Figure 3.14 Distribution of sample weights, WT ...................................................................... 28 Figure 3.15 Distribution of BER ratings in RTB data before and after adjustments ................ 29 Figure 3.16 Distribution of BER ratings in SEAI data before and after adjustments ............... 30 Figure 3.17 Estimated rental housing stock by BER ................................................................ 31 Figure 4.1 Key characteristics by upgrade expenditures dataset .......................................... 35 Figure 4.2 Summary of BER distributions in expenditure datasets ....................................... 37 Figure 4.3 Estimated average costs of upgrade by pre-upgrade ber in €1k in 2023 prices .. 41 Figure 4.4 Cumulative costs and number of dwellings to upgrade estimates ...................... 45 Figure 5.1 Proportion of households that own other properties apart from main residence .............................................................................................................................. 49 Figure 5.2 Per cent of households owning other property by year ...................................... 49 Figure 5.3 Comparing landlords and non-landlords: Age ...................................................... 50 2|Investment requirements for energy efficiency upgrades in the rental sector the investment cost would be to upgrade the stock to a higher energy efficiency level. Finally, we explore whether household landlords, 1 who make up a large proportion of the owners in the sector, have the financial capacity to make these investments in energy efficiency. More specifically, we attempt to answer the following questions: • What is the current energy-efficiency profile of the rental sector in terms of the dwelling types, locations and properties? • What level(s) of investment would be required to increase the above housing stock to more energy efficient levels? • Do household landlords have the financial capacity to make investments in energy efficiency? For the final question, our focus is on household landlords for two reasons: first, they make up a large proportion of the market; and second, they are more likely to face barriers to investment relative to large institutional landlords. For example, larger, commercial landlords face different financing conditions and financial capacity than smaller household landlords. They also face differing incentives around the payback period and rate of return on any residential investment activity. However, the rental sector has many household landlords who have few properties and may not have sufficient wealth to finance the upgrade plans. Understanding the profile and structure of these landlords is critically important in terms of understanding the sector’s investment outlook and capacity. Our assessment of the investment capacity of these landlords is done from a financial perspective and does not take into consideration the availability of existing policy supports of which they can avail. Addressing these questions is hindered by several notable data gaps. First, there is no national database of all rental properties that would include verified BER certificates and would provide detailed information on the various property characteristics needed to explore any potential upgrades. Second, information is required on the typical costs of energy efficiency upgrades and the corresponding change in the BER. To bridge these data gaps, we draw from a number of different datasets both as core analytical tools and also as secondary robustness checks. To profile the energy efficiency of the sector, we firstly use data from the annual and new tenancies registrations from the Residential Tenancies Board (RTB) for the 12-month period of April 2022 to April 2023. Information on new tenancies has been collected since 2007. However, annual registration of all tenancies in Ireland has been a legal requirement since April 2022. More details on this dataset can be found in 1 A household landlord is defined as a household that owns a residential property other than their main household residence and rents out or leases that property. Note this does not include institutional landlords. Introduction|3 Slaymaker and Shiel (2023). In this dataset, we obtain self-reported BER certificate values, which are available for approximately half of the properties. The overall dataset size is approximately 190,000 observations. 2 The dataset also contains selected information on the dwellings such as their type (semi-detached, detached, terrace, apartment) as well as floor area and location. As these data only contain a self-reported BER, we require a robustness check to cross-examine the self-reported information. For these purposes, we also draw on the Sustainable Energy Authority of Ireland (SEAI) property-level BER Research Tool database, which contains a large range of information on the energy efficiency of BER-assessed properties. From these data, a subset can be extracted of the properties that have had a BER assessment for the purposes of renting the property on the private market. For both the RTB and SEAI datasets, we move from the sample to a population estimate by taking a set of weights by county and property type from the Irish Census of Population 2022. Additionally, we estimate the number of rental properties in protected heritage buildings. These buildings are exempt from BER ratings and are excluded from the cost estimates. To understand the cost of upgrading individual properties in Ireland, we combine two datasets. First, a database from the Department of Housing, Local Government and Heritage contains information on the investment expenditures on upgrading properties owned by the local authorities as part of their social housing stock. This Local Authorities Social Housing Upgrade (LASHU) dataset contains detailed information on the property, including housing type, information on the upgrade and, critically, the BER ratings before and after the upgrade. Our dataset contains 390 upgrades across 16 local authorities for the year 2022. Second, we use a dataset, provided by the SEAI, that includes a sample of properties that have received grants for the efficiency upgrades from the SEAI’s One Stop Shop (OSS) services. This dataset includes 1,068 upgrades by both privately-owned properties and properties of the approved housing bodies (AHBs). From the combined LASHU and OSS data, we estimate the average cost of efficiency upgrades per dwelling with multiple regression models. We use the postupgrade B ratings as the upgrade scenario. The upgrade costs are then modelled as a function of pre-upgrade BER, location, dwelling type and dwelling size. The aggregate cost estimates are obtained by combining per-dwelling costs estimates with the population estimates of the rental housings stock. Finally, our assessment of the financial capacity of household landlords is undertaken using the Central Statistic Offices (CSO) Household Finance and Consumption Survey (HFCS). 2 There are a number of reasons why landlords may not report their BER, such as lack of awareness of the information, accidental missing data or possible other strategic reasons. We do not have any data that provides insight into these factors at present. These data are described in more detail in Section 3. 4|Investment requirements for energy efficiency upgrades in the rental sector The paper is structured as follows. Chapter 2 presents the international literature and Irish context. Chapter 3 profiles the energy efficiency of the sector. Chapter 4 considers investment requirements for upgrades. Chapter 5 explores the financial capacity of household landlords. Chapter 6 concludes. Background and context|5 CHAPTER 2 Background and context 2.1 THE IRISH HOUSING MARKET In line with the challenges in the broader housing market in Ireland, the private rental sector (PRS) has been providing a growing share of housing in the past number of years. Figure 2.1 presents the tenure structure of Irish households from the Census from 1991 onwards. The figure presents three groups of households: those in local authority housing; private renting households, which include rentals from approved housing bodies (AHBs); and owner-occupied housing. The share of owner-occupied housing has declined from a high of 82 per cent in 1991 to 70 per cent in 2022. This decline has seen a corresponding increase in the share of private rental properties. Two further insights from the Census data are important in terms of contextualising the changing role of the PRS: Figure 2.1 also shows the proportion of households with children, overall and for households living in the PRS, as well as the proportion of households in the PRS across the age distribution of the household. The largest increases in the proportion of renters across the age distribution are occurring among those in the ‘family formation’ age group (30–44 years). The rate is also increasing among older renters (65+ years). Furthermore, the proportion of households with children in the rental sector increased between 2011 and 2022. This changing demographic structure may present greater challenges in terms of the energy efficiency requirements of the dwellings over time. 6|Investment requirements for energy efficiency upgrades in the rental sector FIGURE 2.1 TENURE STRUCTURE OF IRISH HOUSEHOLDS FROM CENSUS DATA A: Overall tenure structure B: Proportion of households with children C: Age distribution of renters Source: CSO Census data. Notes: Owner-occupied housing includes those owned either by mortgage or outright. Dwellings occupied free of rent and those for whom ‘not stated’ is recorded regarding the nature of the occupancy are excluded. 82% 81% 78% 72% 71% 70% 8% 12% 14% 20% 20% 21% 10% 7% 8% 8% 9% 9% 0% 20% 40% 60% 80% 100% 1991 2002 2006 2011 2016 2022 Local auhorities rentals Private and AHB rentals Owner-occupied 47% 47% 45% 35% 41% 37% 0% 20% 40% 60% 2011 2016 2022 All types of occupancy Private rental sector 0% 20% 40% 60% 80% Under 25 years 25 - 29 years 30 - 34 years 35 - 39 years 40 - 44 years 45 - 49 years 50 - 54 years 55 - 59 years 60 - 64 years 65 years and over 2011 Census 2016 Census 2022 Census Background and context|7 A number of research papers have studied these dynamics. McQuinn et al. (2021) and Slaymaker et al. (2022) have both noted that the challenges in terms of homeownership have been due to house prices outstripping income growth for young households and housing supply remaining well below the level needed for standstill household formation. Indeed, Slaymaker et al. (2022) indicate that a structurally lower rate of homeownership is likely to continue, with more households remaining in rented accommodation throughout their lifecycle. From the perspective of this research, the critical point is that a greater share of the housing stock is now for rental accommodation. For this reason, the challenge of managing the split incentive in an energy efficiency context is even greater than it would have been historically. A further complication arises due to the quality of the housing stock in the rental sector in Ireland. The proportion of properties with low energy efficiency and a very high retrofit requirement is higher than that for the owner-occupied sector (Petrov and Ryan, 2021). This is due to the nature of the housing stock in the PRS, which tends to be older and to have a lower building energy rating (BER). Figure 2.2 presents the structure of the BER ratings for properties in different groups of the rental sector from the Central Statistics Office (CSO) report, Rental market in Ireland 2021. These data split the sector into different groups of the market based on BER status and the type of landlord. It must be noted that these data are taken from the Residential Tenancies Board (RTB) residential tenancies registration dataset, which at that time covered only new registrations and Part 4 renewals. It does not therefore provide an assessment of the entirety of the rental housing stock; rather, it oversamples properties new to the market (including new construction) and those that turn over on a regular basis. The groupings presented are: approved housing bodies (AHBs), Housing Assistance Payment (HAP) recipients, Rent Supplement recipients, 3 local authority properties, PRS housing with private individual household landlords and PRS housing with nonhousehold owners (such as investment funds and other institutional landlords). It is very clear that the individual landlord owned properties, as well as those inhabited by HAP and Rent Supplement recipients, who also live in PRS accommodation, have the highest share of low BER properties; at least 50 per cent of the properties in these groupings are below a C rating. Both AHB properties and those in the non-household PRS sector have a higher share of A–C BER-rated properties, mainly due to the fact that this housing stock tends to be newer. 3 Note both HAP and Rent Supplement recipients live in PRS housing. 8|Investment requirements for energy efficiency upgrades in the rental sector FIGURE 2.2 BER STRUCTURES FOR DIFFERENT GROUPS OF RENTAL SECTOR (% OF TOTAL) Source: CSO data for 2020. Notes: The RTB data only relate to new tenancies and Part 4 renewals, as the RTB did not collect annual registrations for the period in which the CSO undertook this analysis. PRS=Private rental sector. Note both HAP and Rent Supplement recipients live in private rental sector housing. The category of ‘individual landlord’ used here refers to those landlords who registered with the RTB using a PPS number, while the non-household landlords used a company registration office number. This can be seen more clearly in Figure 2.3, which presents the age of properties owned by the same groupings of landlords. The majority of properties owned by the AHB sector as well as the non-household PRS sector were built post 2000, whereas this share is much lower for individual PRS providers as well as the local authority housing stock. FIGURE 2.3 PROPERTY AGE BY LANDLORD TYPE (% OF TOTAL) Source: CSO data for 2020. Notes: The RTB data only relate to new tenancies and Part 4 renewals, as the RTB did not collect annual registrations for the period in which the CSO undertook this analysis. PRS=Private rental sector. Note both HAP and Rent Supplement recipients live in PRS housing. 13 11110 2 28 812 9 22 10 34 40 46 40 34 36 19 40 34 40 27 39 612 710 713 0 20 40 60 80 100 AHB HAP Local authorities PRS individual landlords PRS nonhousehold landlords Rent Supplement housing A B C D-E F-G 171757 3 14 13 12 613 6 15 38 15 5 15 24 21 16 23 13 22 54 43 31 42 60 41 12 11111 2 0 20 40 60 80 100 AHB HAP Local authorities PRS individual landlords PRS nonhousehold landlords Rent Supplement housing Before 1919 1919 to 1970 1971 to 1990 1991 to 2000 2001 to 2010 2011 or later Background and context|9 These data indicate a considerable investment challenge for the sector, and in particular for individual landlords, if energy efficiency commitments are going to be met by the sector. 2.2 INTERNATIONAL LITERATURE The issue of investment in energy efficiency technologies has come to the fore in recent years in line with aims to transition to low carbon economies internationally. Existing research indicates a general ‘energy-efficiency gap’, whereby the economic level of investment suggested by cost minimising (or energy saving) levels is well below that which is actually undertaken by households and firms. Allcott and Greenstone (2012) provide a detailed discussion of this issue and note that the ‘win–win’ argument for investment in energy-saving technology is that it can save fossil fuels (thus reducing all the harmful externalities that come from their usage) as well as help bridge an inefficient level of market investment by participants. These two concepts intertwine two sets of market failure, which are important to separate out when trying to understand the underinvestment in energy efficiency. The first is the issue of negative externalities of the production of fossil fuels; the second concerns our understanding of the barriers to investment and the extent to which information asymmetries or other frictions such as credit market imperfections are driving investment choices. Allcott and Greenstone (2012) note the policy response varies depending on the two market failures; Pigouvian taxes or cap and trade programmes can be used for externalities, whereas other instruments to subsidise or mandate energy efficiency can be used to address the underinvestment. In the context of this particular research, our focus is on the market failures relating to underinvestment in the residential real estate market. The types of market failures that can occur in this case are noted by Allcott and Greenstone (2012): information imperfections (a lack of information on what the optimal level of investment is for an individual household); inattention (missing key elements of the choice decision during the purchase decision); credit market access; and moral hazard. Given these factors, Allcott and Greenstone (2012) note that there is likely a very differentiated heterogeneity in investment inefficiencies across the population, thus policy targeting is required to deal with the differentiated challenges. However, challenges have been found in designing and implementing these policies internationally, such as attempting to use targeted instruments that do not have the desired impact (for example, Murphy et al., (2012) document these issues for the Netherlands by noting that the policies do not take into consideration the complexity involved with regard to existing dwellings). These investment inefficiencies are all the more acute in the rental side of the housing sector due to the ‘split incentive’ problem. This issue relates to the 10|Investment requirements for energy efficiency upgrades in the rental sector situation whereby landlords are the investors, but the tenants are the ones who reap the reward through lower energy bills or other energy-saving benefits. This split incentive makes all the above market failures more acute and challenging to overcome. Castellazzi et al. (2017) note four specific types of split incentives: • efficiency-related split incentives (the tenant pays the electricity bills but cannot choose the technology to improve the efficiency); • usage-related split incentives (when occupants are not responsible for paying their utility bills and therefore have little or no interest to conserve energy); • multi-tenant, multi-owner split incentives (this occurs where consensus is required for energy efficiency upgrades amongst a heterogeneous group of tenants/owners); and • temporal split incentives (where the energy efficiency investment will not pay off before the property gets transferred across ownership). Some research has found that energy performance certificates can mitigate some of these issues. Dwellings with higher levels of energy efficiency have a higher sales value, as well as a higher rental value (Fuerst et al., 2020). However, Cornago and Dressler (2020) document that landlords do not always disclose the energy certificates to tenants even if the certificate exist, and that many prospective tenants do not properly account for energy costs when deciding on which property to rent. Ástmarsson et al. (2013) note that this misalignment of interests is one of the greatest barriers hindering the investment in sustainability from an energy efficiency perspective in residential buildings in Europe. A voluminous literature explores this issue internationally. A recent systematic review of the literature by Lang et al. (2021) notes the poorer energy efficiency of rented homes to owneroccupied properties in many countries across Europe, North America and Australasia. They note that small-scale landlords are the key decision makers and very little is known about their decisions. Looking across 16 papers, they find that 47 factors have been noted as determining their behaviour, including financial factors, values, beliefs, property-market factors and other aspects of their relationships with tenants. Nie et al. (2020) explore the adoption of energy-saving measures between homeowners and renters in a survey of 1,248 households across three countries (Germany, Netherlands and Belgium). They find clear evidence of split incentive problems in relation to both energy efficient technology adoption and energysaving behaviours. They find that homeowners are 16 per cent more likely than renters to adopt these technologies, though with a lower difference regarding behavioural measures. Background and context|11 Some of the reasons for non-investment in energy efficiency by landlords are noted in a paper on the UK market by Hope and Booth (2014). They study the reasons for landlords choosing not to invest in energy efficiency technologies, finding that the majority (67 per cent) indicate ‘high-upfront costs’; other notable reported factors include ‘tenants are happy with the energy efficiency’ and ‘no personal benefit to making improvements’ (40 per cent). Access to finance or lack of information were not noted as barriers in their research. These findings are also echoed in research by Ambrose (2015) who undertook a research interview with 30 landlords in northern England and identified these relevant issues: split incentives, time costs, burden and information on the options. Further research by Miu and Hawkins (2020) surveys the retrofit behaviour of private landlords in the UK and assesses their engagement across 18 different energy efficiency measures. They group landlords into seven behavioural typologies or landlord retrofitters, and suggest a segmentation of the landlord population into different target groups for heterogeneous policy interventions. They note that tailoring policy can better deal with a number of issues including policy support take-up, increasing the likelihood of retrofit and accelerating the energy-efficiency transition. Further evidence is also available to support policy instrument combinations to deal with this issue. In research on the Danish rental sector, Ástmarsson et al. (2013) find that these principal agent problems can only be overcome with a package solution that includes legislative changes, financial incentives and better dissemination of information. In an attempt to provide a cross-country solution to the informational asymmetries component of the energy efficiency gap in rental housing, a major EU-funded research project RentCal produced a tool that can help break down information barriers (Zeitler, 2018). Other studies look at different aspects of the regulations used to incentivise investments in energy efficient technologies. For Germany, Weber and Wolff (2018) find that landlords pass on investment costs to tenants as is allowable under rent control legislation, and these costs are higher than the energy efficiency savings. This is an important finding in an Irish context as such an exemption is allowable in Rent Pressure Zone areas. Charlier (2015), in a study on French data, shows tenants are lower income and unable to invest due to insufficient funds. Maruejols and Young (2011) use Canadian data and find that tenants’ behaviour depends on whether they face the cost of energy usage amounts. 18|Profiling the energy efficiency of the private rental sector Figure 3.5 presents the distribution of BER ratings across those properties listed as having a self-reported BER. It shows that very few properties in the Irish PRS have an A rating; just over 10 per cent of the properties are listed as having an A rating, with fewer than 1 per cent having an A1 rating. In terms of B-rated PRS properties, 8 per cent have a B3 rating, 4.6 per cent have a B2 rating and 2.9 per cent have a B1 rating, totalling 15.5 per cent of properties with an overall B rating. As the energy efficiency requirements are likely to encourage dwellings to be at least B rated, these data highlight the considerable challenge facing the sector in terms of investing sufficiently to reach this particular level. Indeed, according to these data, just under three in every four properties in the rental sector do not meet a B rating. A majority of properties have either a C or D rating; 13 per cent have a C1 rating, 12 per cent have a C2 rating, and 14 per cent have a C3 rating, totalling 38 per cent. Regarding D-rated properties, 12.6 per cent have a D1 rating, while 9.6 per cent have a D2 rating. Focusing in on the lowest rated properties, which are likely to have the greatest challenge in terms of the energy efficiency investment requirements, 8.2 per cent have an E rating, 2.9 per cent have an F rating and 3 per cent have a G rating – the lowest possible BER. 6 FIGURE 3. 5 DISTRIBUTION OF SELF-REPORTED BER OF RTB-REGISTERED PROPERTIES Source: RTB Registrations data. Note: Excluding properties without self-reported BER. Data without adjustments. In order to provide more granular detail on which properties have different BER ratings and where those properties are located, Figure 3.6 presents high-level BER distributions across property types and across counties. Focusing on the geographic split, data are presented for Dublin, Cork, Galway, Limerick, Waterford and ‘Other counties combined’. It is clear that more of the A-rated properties are located in Dublin; this likely reflects the fact that in recent years Dublin has accounted for a greater proportion of new housing supply in the rental sector, 6 For any further information on the BER scale etc, please see: https://www.seai.ie/publications/Your-Guide-toBuilding-Energy-Rating.pdf. Investment requirements for energy efficiency upgrades in the rental sector|19 many of which are new, build-to-rent properties (Daly, 2023). These newly constructed properties will have been built under the current higher energy rating standards. Cork has the second highest share of Bor higher rated properties after Dublin. Galway and Limerick are the areas with the greatest proportion of D or lower ratings in the data. Figure 3.6 also presents the high level ratings by property type: apartment, detached, semi-detached and terraced. Apartments represent the most energy efficient group, with the highest share of Aor B-rated properties. Houses, of any type, had fewer than 20 per cent of the stock at B or higher rating but approximately 60 per cent across these groups had a C or higher rating. FIGURE 3.6 BER DISTRIBUTION BY COUNTY AND DWELLING TYPE County Dwelling type Source: RTB Registrations data. 3.3 SEAI BER DATABASE The RTB dataset described above is the largest sample available on the rental sector at the micro level, with energy efficiency indicators. However, due to the self-reported nature of the information and the large proportion of non-reported ratings, there may be some biases in the information, whereby landlords may misreport the true rating or where the data may be missing systematically. To attempt to provide a robustness check against this occurrence, we draw on a second data source: the SEAI BER Research Tool micro database, which is made available by the SEAI for research purposes. While these data do not contain a specific indicator for whether a property is currently being rented, or provide a stock of rental market properties, they do have some useful information that we can draw on. These data allow us to identify those properties for which the purpose of the BER certificate application was for ‘private letting’. We assume these properties are active and in the rental sector. It also pools all data across the years of the BER (2009–2023 in our sample). 20|Profiling the energy efficiency of the private rental sector FIGURE 3. 7 NUMBER OF ASSESSMENTS BY ASSESSMENT PURPOSE Source: SEAI BER Research data. The database provided by SEAI contains extensive information that was captured as part of the BER process. The research tool provides data on the BER scheme for approximately 1.1mn observations. It includes all information collected as part of the BER process: energy performance of the dwelling; 7 heating; ventilation; lighting; and property characteristics, etc. The data are anonymised; for example, the meter point reference number (MPRN), name(s) and address have all been removed from each entry. Critically for the purposes of our research, a number of relevant fields (the BER notwithstanding) are included. These include: year of construction, type of property, purpose of BER certificate (sale, rent etc.), and year of application. As noted above, using these data, we can identify a subset of 82,299 observations, from the overall database, that relate to rental properties only. These are the properties whose declared purpose was that the BER was obtained for private letting (i.e. the BER was applied for because the property was to become part of the PRS). The reasons property owners gave for seeking a BER certificate are presented in Figure 3.7. It is clear the vast majority of the BER ratings were obtained for other purposes (for example sale, grant support, owner occupation, etc). A number of points are worth noting. Properties that are currently in the rental sector could have obtained a BER certificate through a sale process, or from a grant application etc. Only using the group of properties that sought a BER rating for the specific purpose of renting the property in our sample means we exclude 7 The BER certificate provides a measured scale of A–G, which gives an energy performance score that is comparable across properties (with A being the highest energy efficiency). Each property is provided a score of energy use per unit floor area per year (kWh/m2/yr). For an example, please see: https://www.seai.ie/home-energy/buildingenergy-rating-ber/understand-a-ber-rating/Sample-BER-Cert.pdf. Investment requirements for energy efficiency upgrades in the rental sector|21 these groups. However, this was unavoidable, as we are unable to identify rental sector properties from within the other categories. There is a second important consideration. Properties could have been in the owner-occupied market and then transferred to the rental sector or from the rental sector to owner occupation. We therefore cannot determine whether or not these 82,299 properties are still in the rental sector at present. It must also be noted that a single property could have multiple BER assessments, in which case it would therefore appear multiple times in the data. Despite these limitations, we use these data as a robustness check on the RTB data, which do not suffer from these entry and exit challenges. FIGURE 3.8 NUMBER OF ASSESSMENTS BY YEAR OF ASSESSMENT Source: SEAI BER Research data. The year of completion for the BER certificates are presented in Figure 3.8. It shows that the privately let properties in the sample had their BER assessment completed at different points in time. The figure covers privately let properties and all other purposes combined. While overall a greater proportion of BER certificates have been obtained in more recent years, in the privately-let sample, more than onethird of assessments are from 2014 or earlier. Figure 3.9 presents properties for private rental and other purposes across a number of housing stock categories: apartments, detached houses, semi-detached houses, terraced houses and other. The private lettings data are much more skewed towards apartments, with just over 50 per cent of the observations coming from this housing type. There are fewer detached and semi-detached houses in the rental sample compared to the ‘other purposes’ sample. 22|Profiling the energy efficiency of the private rental sector FIGURE 3. 9 NUMBER OF ASSESSMENTS BY DWELLING TYPE Source: SEAI BER Research data. FIGURE 3.10 NUMBER OF ASSESSMENTS BY YEAR OF CONSTRUCTION Source: SEAI BER Research data. An interesting factor available from the SEAI data that is not available in the RTB data is property age. Older properties are likely to be of poor quality regarding energy efficiency, if they have not been upgraded. Therefore, this is an important variable in terms of providing insight into our understanding of the investment requirements for the sector. Figure 3.10 shows the age distribution of privately rented and other properties by BER status. Two interesting trends emerge: there are more very old properties (pre-1900) in the rental sector; and fewer privatelylet properties were built during the 1960s, 1970s and 1980s. That period (1960s to Investment requirements for energy efficiency upgrades in the rental sector|23 1980s) saw a major expansion in homeownership in Ireland; many of the new builds from that era are likely to have remained in that tenure category. By contrast, the 2000s saw a greater proportion of privately-let properties being built, as it was during that decade that buy-to-lets became a major part of the Irish housing market. FIGURE 3.11 NUMBER OF ASSESSMENTS BY FLOOR AREA Source: SEAI BER Research data. Figure 3.11 shows the size distribution for private rental BER ratings and the rest of the dataset. The metric presented is the floor area in metres squared. Two overlaid histograms are presented, with the blue data representing the rental sector. These data indicate that the properties in the rental sector are typically smaller than their equivalents in the other categories. FIGURE 3.12 BER RATINGS DISTRIBUTION OF ASSESSMENTS Source: SEAI BER research data. 24|Profiling the energy efficiency of the private rental sector Finally, and of critical importance, is the BER distribution associated with these data. This is presented in Figure 3.12. It shows there are disproportionately more C-, Dand E-rated properties in the rental sector data, with notably fewer A rated properties. One possible reason for the overall lower energy efficiency in the SEAI data, compared to the RTB data, is that the SEAI sample includes historic data. Figure 3.13 shows the change in the distribution of BER ratings over the years of assessment. There is a notable increase in energy efficiency in both the rental sector and in the ‘other purposes’ groups. The figure also shows that the rental sector’s energy efficiency is lagging behind that of buildings assessed for other purposes. FIGURE 3.13 BER ASSESSMENTS DISTRIBUTION BY YEAR OF ASSESSMENT Private lettings Other purposes Source: SEAI BER Research data. Investment requirements for energy efficiency upgrades in the rental sector|25 3.4 CENSUS RECONCILIATION AND COMPARATIVE ANALYSIS Having reviewed both the RTB and SEAI datasets, our next goal is to estimate the total number of dwellings in the PRS at each BER level. In doing so, we also make adjustments to the data to account for protected heritage buildings. These buildings are BER exempt and restrictions apply regarding the types of potential energy efficiency upgrades that could be carried out on them, which would likely impact the upgrade costs. These protected buildings are therefore outside of the scope of this report, which uses the current National Retrofit Plan, published as part of Climate Action Plan 2021, as its baseline context. The details of this analysis are presented in Appendix B. 3.4.1 Residential Tenancies Board data As shown in Figure 3.4, 52 per cent of observations in our RTB dataset do not include a BER rating. One of the challenges here is the potential for bias in the selfreported BER distribution. For example, it is possible that some landlords with less energy-efficient properties may not report their BER rating, which would bias our sample distribution towards having a higher rating than is the case for the actual population of properties in the sector. Furthermore, there could be impacts of bias whereby those properties complying with RTB registration in the first place may be more likely to have a high BER and to report it. There are also likely to be other confounding effects that can impact the distribution of self-reported BER ratings, which are not listed here (such as economic or legal variables that impact the preference of the landlord for compliance with the registration process). Two biases are therefore worth considering. The first is whether the data that includes self-reported BER ratings, in the RTB data, are systematically different from those which do not. The second is whether the RTB sample is representative of the overall population of rental properties. To explore the first issue, we present a number of tables that compare properties with a self-reported BER rating against those that do not have a self-reported BER. If any major systematic differences are found to exist between the two groups, this would support the possibility of bias in the BER reporting. The first set of characteristics considered in our assessment are as follows: floor area; monthly rent; number of tenants; number of bedrooms; and property type. The data are presented in Table 3.2. The observations for ‘no BER’ have lower rent and are also smaller in terms of floor area, number of tenants and number of bedrooms. There are proportionally more detached houses without a reported BER than with one (10.3 per cent to 9 per cent). There is a higher share of apartments with a reported BER, with apartments making up 51.6 per cent of the sample with a BER rating, compared to 50.5 per cent of the without one. The proportions of semi-detached and terrace houses are similar in both samples, with no statistically significant difference. 26|Profiling the energy efficiency of the private rental sector TABLE 3.2 CHARACTERISTICS OF PROPERTIES WITH BER COMPARED TO WITHOUT BER Variable No BER With BER Difference Floor area 88.92 89.6 -0.68*** Monthly rent 1302.2 1489.8 -187.7*** Number of tenants 1.828 1.886 -0.058*** Number of bedrooms 2.416 2.475 -0.059*** Apartments 0.505 0.516 -0.011*** Detached 0.103 0.090 0.014*** Semi-detached 0.235 0.238 -0.003 Terraced 0.157 0.157 0.000 Source: RTB Registrations microdata. Notes: *** significant at 1 per cent level using t-test. Floor area trimmed 5, 95 per cent for outliers. We now consider the differences between properties with reported BER status versus those without this on a geographic basis. These are presented in Table 3.3. In Dublin, there is a notably higher proportion of properties with a reported BER status (Dublin makes up 46 per cent of the total ‘with BER’ sample) than without (Dublin properties comprise 40 per cent of the ‘without BER’ sample) compared to the breakdown in other areas. In Cork, there is a higher proportion of properties without a reported BER status (Cork accounts for 12.3 per cent of the total ‘without BER’ sample) than properties with one (Cork makes up 9.9 per cent of the total ‘with BER’ sample). There are also differences in the other areas presented, with Galway making up a comparatively higher share of the ‘with BER’ sample and Limerick, Waterford and the rest of the country accounting for a comparatively higher share of the ‘without BER’ sample. TABLE 3.3 CHARACTERISTICS OF PROPERTIES WITH BER COMPARED TO WITHOUT BER County Without BER With BER Difference Co. Dublin 0.406 0.461 -0.055*** Co. Cork 0.123 0.099 0.024*** Co. Galway 0.054 0.061 -0.008*** Co. Limerick 0.045 0.031 0.014*** Co. Waterford 0.026 0.020 0.006*** Rest of the country 0.346 0.327 0.019*** Source: RTB Registrations microdata. Notes: *** significant at 1 per cent level using t-test/ Floor area trimmed 5, 95 per cent for outliers. Tables 3.2 and 3.3 clearly show that differences exist between the properties which have and do not have a self-reported BER rating, based on observable Investment requirements for energy efficiency upgrades in the rental sector|27 characteristics. For this reason, we propose the following methodology to deal with this issue, based on developing a set of probability weights. We first define a dummy variable which takes the value of 1 for those properties which have a selfreported BER, and 0 otherwise: 𝐻𝑎𝑠𝐵𝐸𝑅 = { 1 if BER reported 0 otherwise We then run a regression model of the probability of not having a BER rating as a function of observable characteristics. In our list of observable characteristics, we include the following: the floor area and rent amount as levels and their squared terms, property type dummies, and indicator variables for the county: Pr(HasBER=0)=𝑓(rent ,rent2, floor ,floor2 , dwelling type , urban , county) This probability is estimated as a logit model, with the results shown in the appendix. 8 Following the estimation, we predict for each property the probability of having a self-reported BER 𝑝𝑖 based on the characteristics in the regression and the estimated coefficients. We then use these predicted probabilities to re-weight the sample. The resulting distribution is shown in the middle columns in Figure 3.15. In the reweighted sample, the proportion of A* and B* ratings is lower than before the adjustment, while increases are seen for C* and D* ratings. The proportion of properties rated E and below remains similar. A final sample adjustment that we make is to further re-weight the RTB sample by county and dwelling type, such that the number of observations in the RTB sample corresponds to the Census 2022 data by county and dwelling type (of which there were 330,632 dwellings in the PRS). We attempt to match the data as closely as possible in terms of dwelling types but common groupings are required. The mapping that corresponds the RTB data and the Census data is presented in Table A.1. Furthermore, we make a number of adjustments to account for the BER-exempt status of the protected heritage buildings in the RTB data. First, we develop a process that attempts to identify listed buildings in the data and to remove these from our analysis of energy upgrade requirements. This process is outlined in Appendix B, and leads to approximately 5 per cent of the RTB observations being identified as of a protected nature, thus reducing the RTB sample from 209,035 to 196,305 dwellings. Given the special requirements of these buildings in terms of 8 Due to the presence of outliers, monthly rent and floor area have been trimmed for the bottom and top percentiles. Additionally, some observations had missing information in these two variables. For those observations, the values were imputed. A logit model is an estimation procedure that uses a distributional form catering for binary outcome variables. It draws on the logistic distribution. 34|Understanding investment expenditure needs from the LASHU data. The BER ratings provided in Table 4.1 relate to the broad A*, B* and C* ratings. The majority of the upgrades are to A* ratings (56 per cent); however, this is driven primarily by private properties from the OSS dataset. The LASHU dataset has more B* than A* ratings, while for AHBs the upgrades are equally split between A* and B*. TABLE 4.1 NUMBER OF OBSERVATIONS BY DATASET Post-upgrade BER rating Dataset C* B* A* Total OSS AHB 0 355 350 705 OSS private 0 36 327 363 Social housing 9 241 140 390 Total 9 632 817 1,458 Source: LASHU and SEAI. Only observations with non-missing cost of upgrade are included. In terms of the types of properties and their geographic locations in these datasets, some simple descriptive statistics are presented in Figure 4.1. Three variables are included: 1) a Dublin indicator capturing the proportion of properties in Co. Dublin; 2) an apartment indicator capturing the proportion of properties that are apartments; and 3) a large dwelling indicator, which gives proportions for those properties whose floor area is above 100m2. In all three charts, the average value from the Census is provided for reference as the horizontal line. In terms of location, the LASHU dataset is close to both the RTB and SEAI PRS databases, with between 40 and 50 per cent of the properties in Dublin across these datasets. In contrast, the AHB retrofits were primarily conducted outside of Dublin (less than 20 per cent in Dublin) and only just over 30 per cent of the OSS private upgrades were in Dublin. Regarding property type and specifically the share of apartments, the LASHU dataset is closest to the RTB and SEAI private rental datasets, with more than 40 per cent of LASHU properties being apartments compared to 51 per cent in the RTB and SEAI samples. In contrast, the OSS data contain mostly houses rather than apartments. This is unsurprising given that the policy is targeted at homeowners, and houses are likely to be easier to retrofit on average than multi-unit dwellings with common areas. The final variable presented in Figure 4.1 relates to larger properties (defined as 100m2 or above). The OSS private sample has a significantly higher share of large properties, at nearly 50 per cent. This likely reflects the share of houses in the data, which are larger than apartments in general. The RTB has the second highest share of large properties while the SEAI, LASHU and AHB (OSS) samples have a considerably lower share. Investment requirements for energy efficiency upgrades in the rental sector|35 FIGURE 4.1 KEY CHARACTERISTICS BY UPGRADE EXPENDITURES DATASET Source: RTB Registrations microdata, SEAI BER Research data. In general, comparing across these variables, we find that the OSS data for private homeowners are much more likely to concern properties outside of Dublin and larger houses, while the LASHU data in particular is more similar in the share of properties that are in Dublin and that are apartments to the RTB and SEAI private rental sector data outlined in Chapter 3. Figure 4.2 presents three key data fields across the various sub-samples and for the combined cost data. The first chart for each of the sub-samples relates to the preand post-BER ratings distributions. This is a critical piece of information for our research as it plots the observed changes in energy efficiency for the properties for which we have cost data. Focusing first on the social housing upgrades data, it is clear the majority of properties had a very poor BER rating before the upgrade, with the majority of properties having an E or D rating. However, after the upgrades most properties had a B2, B1 or A3 rating; this represents quite a significant increase. For the AHB sample, the quality of the housing stock appears to have been somewhat better as the majority of properties before the upgrade were D1 or C rated. After the upgrades, most properties were B1 or A3 rated. For the OSS private sample, quite a uniform distribution across the ratings from C2 down is evident before the interventions. There were more Gand F-rated properties in this sample than in the other datasets. Following the interventions, the vast majority of the properties that came through the OSS private scheme had an A rating, with approximately 10 per cent having a B rating. This represents quite a major change in terms of energy efficiency. The final panel in Figure 4.2 includes the overall sample, with the majority of the pre-upgrade distributions populated by D-and C-rated properties while the majority of the post-work BER ratings were B2 or A3 (nearly 70 per cent). The second figure (middle column) presented for each of the sub-samples is the distribution of the number of BER changes. The OSS private sample sees the largest jumps in terms of the BER ratings, with many properties moving up 7–12 places on the BER scale. The OSS AHB dataset has more moderate changes with, five-point 36|Understanding investment expenditure needs increases being the most frequent jump. These more moderate rating improvements are likely related to the relatively better starting point for these properties. The LASHU data have a fairly dispersed range of rating improvements but the most frequently occurring rank jump involves nineor ten-scale place increases. Overall, the most frequently observed increases are five to seven points, reflecting the size of the AHB sub-sample as a proportion of the overall dataset. The final set of charts (right-hand column) presented in Figure 4.2 show the distribution of the investment costs associated with the energy efficiency upgrades. While the majority of the data presented in the samples are for properties renovated in 2022, the investment cost data have been transformed to 2023 values by deflating the data in line with the Central Statistics Office’s (CSO) cost index for materials and inputs into the construction sector. For the LASHU data, the majority of the upgrades cost between €20k and €40k per property, but the distribution does have a long tail towards the higher values with some large expenditures. For the AHB data, the majority of the expenditure is again between €20k and €40k per property, with little variation. This likely reflects the smaller range of BER rank increases seen in these data. For the OSS private sample, there is a very large spread in terms of the expenditure, and the average and median are much higher for this sample than for the others. This likely reflects the difference in housing types and the larger houses on average, as well as the bigger ratings increases (more A-rated properties after the upgrades) than the other datasets. As noted previously, it is also possible that local authorities in particular may have benefited from some economies of scale through bulk upgrades across multiple properties that individual homeowners in the OSS private sample would not have had. Finally, it cannot be excluded that some homeowners in the OSS private sample may also have included costs for some non-energy efficiency-related expenditures incurred when the works were carried out. Investment requirements for energy efficiency upgrades in the rental sector|37 FIGURE 4.2 SUMMARY OF BER DISTRIBUTIONS IN EXPENDITURE DATASETS Source: LASHU and SEAI One Stop Shot datasets. Note: Total spent graphs do not show outlier values above €100k. 4.3 ESTIMATION OF UPGRADE COST PER DWELLING Having profiled the energy efficiency upgrade cost datasets, our next aim is to estimate a dwelling-specific energy efficiency upgrade cost. To do so, we combine the LASHU and OSS datasets outlined above, and harmonise them to give a consistent classification of dwelling types according to Table A.1 (in the appendix). In line with the national policy target of upgrading the housing stock to a mid-B BER level, we use a sub-sample of observations that had upgrades to either B3, B2 38|Understanding investment expenditure needs or B1, and had a pre-upgrade rating of C or lower. This leaves a regression sample of 631 observations, of which three-quarters (481 obs.) are upgrades to B1, while 123 are upgrades to B2 and the remaining 27 are upgrades to B3. Therefore, the predicted costs of upgrade costs lean towards the higher end of the B rating. The regression sample contains only 36 observations from the privately owned OSS sub-sample, because the majority of these upgrades were to A* level and are therefore omitted from our estimation. The dependent variable is always the log of total costs of the upgrade in 2023 prices. When results are reported (e.g. in Figure 4.3), the logarithmic values are converted back into euros. It is important to note that the regressions are used to estimate the average cost of the energy efficiency upgrade. The actual costs for individual properties will deviate from this expected value – some being lower and some higher. Due to limited data, both in terms of the sample size and observable characteristics, the regression models cannot account for every possible determinant of the upgrade costs. Despite the diversity of the housing stock, when the predicted values are aggregated in Chapter 5, it is likely that these individual differences will tend to counterbalance each other. Consequently, even if individual estimates diverge, the means are still reliable estimates for the aggregate costs. Our first approach is to estimate the costs as a series of dummy variables, where each dummy represents one of the nine pre-upgrade BER ratings, from G to C1. This approach shown in equation Reg.1 is numerically equivalent to estimating average log costs for every pre-upgrade BER rating. The predicted costs for each dummy are shown in Figure 4.3, and the full results are in Table A.4 (appendix). The estimated upgrade costs range from €40k for an F/G rating, to around €26k for a pre-works BER of D or C. 𝑙𝑛𝐶𝑜𝑠𝑡𝑖= 𝛽0 + ∑𝛽𝑗(𝑃𝑟𝑒𝐵𝐸𝑅=𝑗)𝑖+𝜀𝑖 𝐶1 𝑗=𝐻 (Reg.1) 𝑙𝑛𝐶𝑜𝑠𝑡𝑖= 𝛽0 + ∑𝛽𝑗(𝑃𝑟𝑒𝐵𝐸𝑅=𝑗)𝑖 𝐶1 𝑗=𝐻 +𝜷𝑿𝑻+𝜀𝑖 (Reg.2) In Equation Reg.2 additional control variables, represented by vector X, are added to the model. The three control variables are: a) a Dublin dummy, which takes the value 1 if a dwelling is located in County Dublin and 0 otherwise; b) an apartment dummy which equals 1 if the dwelling is an apartment and 0 if the dwelling is a house; 10 and c) a large-sized dummy which equals 1 if the dwelling is larger than 10 See Table A.1 for details on dwelling types across the datasets. Investment requirements for energy efficiency upgrades in the rental sector|39 100m2 and 0 otherwise. All these dwelling characteristics are also reported in the RTB and SEAI datasets, which allows us to predict dwelling-specific costs of the upgrade before aggregating. The results in Table A.4 show that upgrades in Co. Dublin are around 25 per cent more expensive than elsewhere and that apartment upgrades are 20–25 per cent cheaper than houses. The dwellings above 100 m2 are around 15 per cent more expensive to retrofit; however, this relationship is statistically not as strong. The magnitudes of these relationships are similar in all other model specifications. Adding these three variables does not change the average upgrade costs much, though they significantly improve the predictive power of the model. In the remaining four specifications, the relationship between the costs and pre-upgrade BER is modelled as a continuous polynomial function. The 𝑃𝑟𝑒𝐵𝐸𝑅 rating is converted into a numerical variable 𝑃𝑟𝑒𝐵𝐸𝑅  between 0 and 1, with an equally spaced interval between each pre-works BER: 𝑃𝑟𝑒𝐵𝐸𝑅 = { 0if 𝑃𝑟𝑒𝐵𝐸𝑅=𝐻 1/8 if 𝑃𝑟𝑒𝐵𝐸𝑅=𝐹 2/8 if 𝑃𝑟𝑒𝐵𝐸𝑅=𝐸2 ⋮ ⋮ 1if 𝑃𝑟𝑒𝐵𝐸𝑅=𝐶1 In Reg.3 the costs are modelled as a quadratic function and in Reg.4 as a cubic function of this variable. This approach relies on the assumptions that nearby BER ratings will involve similar upgrade costs. This gives better predictions when there are relatively few observations in the pre-upgrade rating, 11 and reduces the risk of overfitting. 𝑙𝑛𝐶𝑜𝑠𝑡𝑖= 𝛽0 + 𝛽1𝑃𝑟𝑒𝐵𝐸𝑅 𝑖 + 𝛽2𝑃𝑟𝑒𝐵𝐸𝑅 𝑖2 +𝜷𝑿𝑻+𝜀𝑖 (Reg.3) 𝑙𝑛𝐶𝑜𝑠𝑡𝑖= 𝛽0 + 𝛽1𝑃𝑟𝑒𝐵𝐸𝑅 𝑖 + 𝛽2𝑃𝑟𝑒𝐵𝐸𝑅 𝑖2 +𝛽3𝑃𝑟𝑒𝐵𝐸𝑅 𝑖3+𝜷𝑿𝑻+𝜀𝑖 (Reg.4) Panels (3) and (4) in Figure 4.3 show the estimated costs, which are broadly in line with results (1) and (2). The costs are higher for upgrades from G at €43.5k as well as for upgrades from D and C, at about €28k. However, the costs for F and E are slightly lower than in regression models (1) and (2) at €32k–38k. To compare the models we calculate the Akaike information criterion (AIC) and the Bayesian information criterion (BIC), which are standard measures of the predictive power of the model. Both the AIC and BIC show the quadratic and cubic equations have better predictive power compared to the series-of-dummies approach. 12 11 For example, there are only 33 observations with a pre-upgrade BER rating of E2. 12 Note that for both AIC and BIC, lower values means better predictive power. 40|Understanding investment expenditure needs The remaining two regression models use equation Reg.3, but are estimated using quantile regression. Model (5) estimates the expected median cost (50th percentile). The predicted values of the medians are again similar to the previous estimates of the mean cost of upgrade. Finally, model (6) estimates the 75th percentile of the upgrade costs as an upper-bound estimate of the upgrade costs. The fitted values are accordingly higher and they range from €32.7k to €49.3k. This scenario can be seen as a useful upper bound, which could occur under a persistent and elevated high construction inflation environment or excessive capacity constraints in the construction sector leading to price increases. In several of the regression models, the estimated relationships are not always strictly decreasing. For example, based on Reg.2 the estimated mean upgrade costs from C1 are €600 higher than an upgrade from the less energy efficient C2 BER rating. This is due to small numbers of observations in the data, especially for C1 as only 12 dwellings in our data got an upgrade from C1 to B*. These nonmonotonic estimates tend to be fairly small and often not statistically significant. Therefore, we do not make any further functional-form assumptions or use nonlinear regression models to address this issue. Investment requirements for energy efficiency upgrades in the rental sector|41 FIGURE 4.3 ESTIMATED AVERAGE COSTS OF UPGRADE BY PRE-UPGRADE BER IN €1K IN 2023 PRICES Source: LASHU and SEAI OSS datasets. Notes: Predicted values at Dublin dummy=0.38, apartment=0.40, large=0.32. 95% confidence intervals. Full estimation results table are in Table A.4 (appendix). 4.4 TOWARDS AN AGGREGATE COST OF PRS DWELLING UPGRADES The next step in our analysis is to develop an aggregate overall cost for the PRS of undertaking energy efficiency upgrades. To do this, we use the combination of datasets and estimates of the cost structures in the previous two chapters. Our general methodology is as follows: for each dataset that measures the BER profile of the sector (RTB and SEAI research database), we have estimated a propertyspecific cost for each dwelling. We then aggregate this cost with the Census 42|Understanding investment expenditure needs weights to obtain a total renovation cost. The estimation of the costs across different methodologies has been outlined in the previous section. As mentioned above, for the purposes of this analysis, we limit ourselves to an upgrade target in line with the National Retrofit Plan, which aims for B2 standard dwellings as a key target. As discussed above, we therefore use the cost data in our estimates to upgrade all properties from their current BER rating to a minimum of the average cost for a B* property in our data. Given the vast majority of our B* upgrade cost data relates to B1 or B2 properties, our upgrade scenario in essence moves all properties to a mix of B2 or B1 levels. These costs are then aggregated across all properties that are currently C rated or below. An illustration of our aggregation process can be seen in Table 4.2. The table draws on the RTB tenancy registrations data sample as described in Chapter 3. The estimates of cost of upgrade are taken from the quadratic fit model (Reg.3) described above. Because this model has the best predictive power it is used as a benchmark model. In Table 4.2, the estimated distribution of current BER ratings, adjusted for nonreported values and Census weighted, is presented in column (2). Column (3) is the proportion of the totals excluding A/B and BER-exempt rental properties. Naturally, for this scenario any property that is already energy efficient does not require an upgrade and will not be included in further calculations. Thus, the total number of properties simulated for upgrade is 242,467. In column (5), the average cost of the upgrade per property is provided for each of the BER groups. For example, the average upgrade costs to B* for all properties is just over €30k but it ranges across the starting BER; the average upgrade cost for G-rated properties is €43k and this declines to €28k for C-rated properties. The total cost for each group is presented in column (6) with the proportion of the total cost in column (8). In this scenario, the total sector upgrade costs are approximately €7.3bn. Just under €5bn of this total relates to properties that are currently D or C rated. Although the average cost is lower than for low-efficient G/F/E rated properties, there are many more mid-efficient properties overall. Investment requirements for energy efficiency upgrades in the rental sector|43 TABLE 4.2 EXAMPLE USING RTB DATA AND QUADRATIC FIT UPGRADE COSTS Dwellings Upgrade cost (1) (2) (3) (4) (5) (6) (7) (8) (9) Current BER Est. number % Cum. % Mean (€1,000s) Total (€mil) Cum. (€mil) Total (%) Cum. (%) G 9,663 4% 4% 43.5 421 421 6% 6% F 9,523 4% 8% 38.9 371 791 5% 11% E2 10,239 4% 12% 34.8 357 1,148 5% 16% E1 16,005 7% 19% 31.7 507 1,655 7% 23% D2 31,136 13% 32% 29.9 932 2,587 13% 35% D1 40,708 17% 48% 28.9 1,176 3,763 16% 51% C3 45,029 19% 67% 28.3 1,274 5,037 17% 69% C2 38,859 16% 83% 28.3 1,100 6,137 15% 84% C1 41,306 17% 100% 28.9 1,194 7,331 16% 100% For upgrade 242,467 100% 100% 30.2 7,331 7,331 100% 100% B* 46,145 Dwellings not included in upgrade cost calculations. A* 23,925 Exempt 18,095 Total 330,632 Notes: Columns 2–4 number of dwellings based on RTB data, adjusted for missing BER ratings, population-weighted with number of dwellings from the Census, and excluding dwellings in protected superstructures (row ‘Exempt’). Column (5) shows average costs in €1k in 2023 price levels, using equation Reg.3 model and combined data from LASHU and the SEAI OSS service. These averages account for dwelling characteristics (apartment, Dublin-based, large-size dummies), and therefore differ slightly from representation in Figure 3.3 where these characteristics are held constant across all pre-upgrade BER ratings. Figures in columns (6) and (7) are in million euros. Table 4.3 presents the range of aggregate estimates calculated across all six models tested in the previous chapters and across both housing stock datasets, the RTB and SEAI data. In all figures and charts, the cost data are provided in 2023 prices and assume upgrade technologies and associated investments costs in line with those in the micro datasets above. The most parsimonious cost equation specification, which does not contain any control variables, gives a total upgrade cost in the RTB data of €6.9bn while the upgrade cost in the SEAI dataset is €7.7bn. When controls (floor area, Dublin and dwelling type) are included, the costs increase to €7.3bn and €7.85bn using the RTB and SEAI datasets respectively. 50|Investment barriers and the landlord structure In terms of the age profile 14 of landlords (Figure 6.3), there is a greater share of residential landlords over the age of 45 (68 per cent) than non-landlords (61 per cent). 15 To provide a more granular split of the data, there are more landlords aged between 46 and 65 years (53 per cent) than non-landlords (36 per cent). FIGURE 5.3 COMPARING LANDLORDS AND NON-LANDLORDS: AGE Source: Analysis based on CSO HFCS data. Data provided by the CSO from their Rental market in Ireland 2021 report show that many of the older landlords have multiple properties (Figure 5.4). The age of the landlord may impact their decisions around investment expenditure on energy efficiency. For example, bank credit access may get more difficult with age, and on retirement, as credit is rationed through a shorter loan term being available and less income to cover repayments; i.e. the households are in that period of their lifecycle in which they are running down accumulated financial assets. Older landlords may have a shorter investment horizon for holding the asset, which may affect the net present value of any investment. The assessment is likely to depend on the cost of the investment, the availability of grants or subsidies (which is outside the scope of this report) and existing BER. It is also possible that the ability to re-price the rent through the Rent Pressure Zone legislation after energy efficiency upgrades would incentivise them to make the expenditure. The degree of compliance and monitoring involved could also factor in their decision-making process. 14 Ages presented refer to heads of household. 15 Ages presented refer to heads of household. 15 Ages presented refer to heads of household. 32 26 27 15 39 20 17 25 0 10 20 30 40 18-45 46-55 56-65 66 plus Landlords Non-landlords Investment requirements for energy efficiency upgrades in the rental sector|51 FIGURE 5.4 LANDLORD AGE BY NUMBER OF TENANCIES Source: CSO data based on RTB tenancy registration data. Notes: The RTB data only relate to new tenancies and Part 4 renewals as the RTB did not collect annual registrations for the period which the CSO undertook the analysis. Finally, we compare the employment status of landlords and non-landlords. Employment status is split into three groups; working, retiree or ‘other’, where other refers to those who are unemployed, on temporary leave, students, in the military, fulfilling domestic services, or permanently disabled (Figure 6.5). Almost one-quarter (23 per cent) of both landlords and non-landlords were found to be retired. More landlords worked (64 per cent) than non-landlords (53 per cent). Nearly double the share of non-landlords were in the other category compared to landlords, at 25 and 13 per cent respectively. 11 1 1 26 11 5 24 53 55 46 53 20 34 49 22 0 20 40 60 80 100 1–2 tenancies 3–19 tenancies 20+ tenancies All landlords Landlords with Per cent Landlord age 65 years and over 45–64 years 30–44 years 0–29 years Number of tenants landlord has 52|Investment barriers and the landlord structure FIGURE 5.5 COMPARING LANDLORDS AND NON-LANDLORDS: EMPLOYMENT Source: Analysis based on CSO HFCS data. 5.1.3 Overview of income and wealth One important aspect of investment capacity is income. We examine both landlord employment income and total earnings – i.e. including income from stocks and bonds, pensions, rent, social transfers and other sources of household income. Table 5.1 below contains the weighted average and median level of income for landlords from the HFCS dataset for the year 2020. The average total household income for landlords was approximately €110,816. The median household income for landlords was €92,400. TABLE 5.1 SUMMARY OF LANDLORD ANNUAL HOUSEHOLD INCOME DATA – SURVEY YEAR 2020 Mean Median Landlords €110,816 €92,400 Source: Analysis based on CSO HFCS data. Another useful aspect to consider in this context is the number of properties owned by landlords across the income distribution. These are presented in Figure 5.6, which draws on the CSO’s Rental market in Ireland 2021 report. 64 13 23 53 25 23 0 20 40 60 80 Working Other Retiree Landlords Non-landlords Investment requirements for energy efficiency upgrades in the rental sector|53 FIGURE 5.6 INCOME DISTRIBUTION OF RTB LANDLORDS BY LANDLORD SIZE Source: CSO data based on RTB tenancy registrations data. These data show that landlords with more than 20 tenancies have considerably higher incomes than those with fewer tenancies. This again highlights the potential financial capacity challenge for lower income, single or ‘few property’ landlords. It is important to note the difference in income distribution across the CSO dataset and the HFCS data. As the CSO only include earned income data, they likely do not capture income from wealth that could enhance the income position of landlords. Furthermore, they only consider new active tenancies as per the RTB dataset; therefore the data are biased towards only considering those tenancies that turn over more frequently. The ESRI/RTB Rent Index (as well as other market monitoring reports such as Daft.ie) show falling turnover in the market, meaning that new tenancies are becoming less representative of the entire market over time. For example, in 2007–2008, over 100,000 new tenancies were registered every year with the RTB. This had dropped to 64,000 in 2021. Another critical factor in the ability of landlords to invest in their properties is the annual return they receive in rent. These cash flows are the yield that could be used to offset and cover any investment expenditures, and are therefore a critical component of any assessment of investment feasibility. Table 5.2 presents the mean and median landlord income from rental properties for the years 2018 and 2020. The average rental income in 2020 was approximately €20k per annum, while the median was €14.4k. These had increased by 14 per cent and 4 per cent respectively over this period. 0 10 20 30 40 50 Per cent of landlords All landlords Landlords with 1–2 tenancies Landlords with 3–19 tenancies Landlords with 20+ tenancies 54|Investment barriers and the landlord structure TABLE 5.2 SUMMARY OF LANDLORD ANNUAL RENTAL INCOME DATA – SURVEY YEARS 2018/2020 Survey year Mean Median 2018 17,627 13,800 2020 20,243 14,400 % Difference +14% +4% Source: Analysis based on CSO HFCS data. Given that landlords of residential properties may combine earned and non-earned income when considering their letting, it is useful to consider what proportion of their total income comes from the rental property. An examination of rental income shows that the mean share of rental income as a proportion of total household income for landlords was 24 per cent, i.e. on average, rental income makes up approximately one-quarter of landlords’ income. Figure 5.7 presents the proportion of landlords against the share of their income that comes from rental income. A majority of landlords earned less than 20 per cent of their income from rental sources. FIGURE 5.7 PROPORTION OF LANDLORDS BY RENTAL INCOME AS % TOTAL INCOME Source: Analysis based on CSO HFCS data. A key factor in terms of the overall ability of landlords to invest in energy efficiency technology is the level of wealth that they hold. This in particular relates to financial wealth or deposits that can be easily deployed for capital investment purposes. In this section, we draw on the data from the HFCS on wealth structures to provide a summary overview of the resources available to landlords. In the HFCS, a number of wealth variables can be derived. Table 5.3 below presents the definitions of the wealth variables that we use in the analysis. 65 59 19 21 17 20 0 20 40 60 80 100 2018 survey 2020 survey % of total household income Over 40% 20%–40% Below 20% Investment requirements for energy efficiency upgrades in the rental sector|55 TABLE 5.3 DEFINITIONS OF WEALTH VARIABLES Variable Definition Real assets Collective value of household main residence, properties, vehicles, self-employment businesses and valuables. Financial assets Collective value of any stocks, bonds, mutual funds, savings accounts, managed accounts, non-self-employment businesses, sight accounts, private lending, voluntary pensions and ‘other’ assets. Total assets Value of real assets and financial assets. Net wealth Total assets – total outstanding balance of a household’s liabilities. Table 5.4 presents the summary statistics of the main wealth variables for landlords. Median total assets of landlords were just over €528k, with a median net wealth of €392.5k. However, this relates to wealth from the principle private residence as well as investments. In terms of financial assets, the median was €31.5k. TABLE 5.4 SUMMARY OF MAIN WEALTH VARIABLES Landlord Variable Mean (€) Median (€) Total assets 922,800 528,094 Net wealth 867,757 392,519 Real assets 772,685 416,100 Financial assets 83,834 31,551 Source: Analysis based on CSO HFCS data. One important aspect to consider when thinking about the issue of household investment is the level of savings. Savings are included above in the ‘financial assets’ category. However, due to liquidity, it is useful to consider them separately. The data are presented below in Table 5.5. 56|Investment barriers and the landlord structure TABLE 5.5 SUMMARY OF SAVINGS (€) Mean (€) Median (€) Landlord 37,148 17,000 Non-landlords 19,746 6,460 Total 21,008 7,000 Source: Analysis based on CSO HFCS data. The mean level of savings for landlord households is €37k and the median is €17k. This suggests that many landlords would have to obtain credit in order to invest substantially in their real estate as the investment costs for their rental properties would likely be much more than this, based on the costs presented in Chapter 4. In addition to their wealth, landlords are often also carrying considerable debts, relating to their investment borrowings but also their own residential dwelling. The data in Table 5.6 present the level of debt and the current loan-to-value ratios of landlords. The mean liabilities carried by landlords were just over €200k in 2020, with a median of €104k. TABLE 5.6 SUMMARY OF INDEBTEDNESS Mean Median Summary of total outstanding Balance on household liabilities (€) 201,138 104,032 Portfolio loan-to-value ratio 37.8% 33.1% Source: Analysis based on CSO HFCS data. Given that it seems a distinct possibility that many landlords would require collateralised credit to finance investment into improvements in their properties, current loan-to-value ratios are examined to explore the extent of collateral available in the properties. These loan-to-value ratios are calculated at the portfolio level, including the main residential dwelling, and include all mortgage loans in the numerator. The mean and median figures are presented in Table 5.6. These landlords have a mean loan-to-value ratio of 38 per cent. This represents the amount of outstanding debt on properties as a proportion of the value of those properties. Therefore, on average, landlords have outstanding debt worth less than 40 per cent of the value of their properties. Hence, it is possible that some landlords would be able to attain financing for improvements. This would depend, however, on the level of investment required, and the term of the loan that would be available to the borrower (likely linked to their age). The issue of financial capacity is discussed below. Investment requirements for energy efficiency upgrades in the rental sector|57 5.2 SIMULATING HYPOTHETICAL FINANCING GAPS Having reviewed the income and wealth position of Ireland’s household landlords, the aim of this section is to get a clearer assessment of the ability, and willingness, of landlords to undertake investments in their properties. In the first exercise we undertake, we use the information on wealth from the HFCS survey and combine this with hypothetical investment expenditures, calibrated from the data in Chapter 4, to simulate in the microdata the extent to which households could cover this expenditure through their own financing resources. Second, for those households that have insufficient resources available, we then explore their ability to finance and cover those expenditures using commercial financing. This can provide insight into the ability of households to bridge the gap. It must be noted that this research does not consider the support available from existing policy mechanisms, which would naturally be available to aid homeowners and landlords. In an attempt to provide some insight into the extent to which Irish landlords have the financial capacity to invest in energy efficiency upgrades, we deploy some simple hypothetical investment scenarios and explore the number of households that could undertake these investments, based on their individual level of wealth and resources as outlined in the HFCS survey. These scenarios are developed to be hypothetical in nature but to reflect the range of investments that could be needed for more straightforward investments, such as attic insulation, and installation for heat pumps towards more complete retrofits. For simplicity, we use four potential investment expenditures of €25k, €50k, €75k and €100k, and simulate how many household landlords could afford to cover these expenditures using their financial assets or deposits in the HFCS micro data. In each of the hypothetical scenarios, we determine a financial gap, which is the difference between the level of the investment and the level of resources that the household has at their disposal: Investment gap = The hypothetical investment level – Total household financial wealth (or deposits) Some households will have sufficient resources to cover the expenditure; therefore, they will not have a gap per se, i.e. the above indicator will be negative. We can therefore define an indicator variable that takes the value of 1 if a household has insufficient resources to cover the expenditure and is 0 otherwise: the investment gap for each J investment value for each household h as follows: 𝐼(𝐺𝐴𝑃)∗𝑗,ℎ={𝐼∗𝑖𝑓 𝐼(𝐺𝐴𝑃)𝑗>0 0𝑖𝑓 𝐼(𝐺𝐴𝑃)𝑗≤0 58|Investment barriers and the landlord structure where: 𝐼(𝐺𝐴𝑃)𝑗= 𝐽−𝐹ℎ 𝐽 takes the values of each of the above investments and 𝐹ℎ is the measure of household ℎ financial wealth used in the scenario. We will therefore report two outcomes from this analysis in our reporting: • share of households with an investment gap (% of landlords); and • the median level of the gap for each household with a gap. Typically, investment is financed by liquid funds or savings held in accounts (if own funds are being used). Therefore, we use two different measures of financial assets in this analysis. First, we use total financial assets, as defined in Section 5.1. However, we also use total deposits as a more realistic indicator of financial resources available to invest, as some landlords may hold other financial assets in longer term, more illiquid holdings, which they may not be able to access (such as pension funds). We present findings for the four hypothetical investment levels as well as the two different measures of financial wealth. Table 5.7 presents the proportion of landlords who are experiencing an investment gap, i.e. the proportion of landlords who are unable to cover the investment expenditures using their financial wealth. The four investment simulations are presented in each row of the table, while the two different measures of the financial resources are presented in the columns. As deposits are included in financial wealth, the strictest scenario in terms of difficulty to achieve for landlords is the high investment (€100k) financed by deposits. TABLE 5.7 PROPORTION OF LANDLORDS UNABLE TO FINANCE THE INVESTMENT ACTIVITY BY FINANCIAL MEASURE AND INVESTMENT SIZE Simulated investment requirement F = Financial wealth F = Deposits J=€25k 49% 49% J=€50k 62% 70% J=€75k 71% 81% J=€100k 77% 86% Source: Analysis based on CSO HFCS data. The results for the proportion of landlords unable to cover the indicative hypothetical investment values using their financial wealth/deposits are presented in Table 5.7. In total, just under half of Irish landlords would have insufficient savings to afford a €25k investment using either their total financial wealth or their savings. This rises to 62 per cent (70 per cent) when considering the use of financial wealth (deposits) to cover the €50k investment. Considering the largest hypothetical investment of €100k, 23 per cent of landlords would have sufficient Investment requirements for energy efficiency upgrades in the rental sector|59 overall financial wealth to cover the expenditure required for this level of investment, while 14 per cent of landlords would have sufficient deposits to cover this expenditure. These figures point to quite a complex picture in terms of the investment capacity of landlords. Nearly half of landlords are unable to cover the smallest investment level considered, of €25k, pointing to a notable absence of investment capital and a low liquid wealth concentration for most landlords. As these landlords are unlikely to be willing to invest all of their funds on energy efficiency, activating retrofits on their properties is going to be challenging without supports. On the other hand, between one-in-five and one-in-seven have sufficient wealth to cover large investments of €100k, depending on whether total financial wealth or deposits are included. To understand more about the investment gap faced by those landlords with insufficient funds, Table 5.8 presents the median investment gap for those landlords, i.e the gap between their own financial assets/deposits and the hypothetical investments. Again, the information is presented in terms of median levels to provide information on the typical gap. TABLE 5.8 MEDIAN INVESTMENT GAP OF LANDLORDS UNABLE TO FINANCE THE INVESTMENT ACTIVITY BY FINANCIAL MEASURE AND INVESTMENT SIZE Simulated investment requirement F = Financial wealth F = Deposits J=€25k 13,966 14,161 J=€50k 32,130 32,384 J=€75k 51,067 51,067 J=€100k 71,197 73,266 Source: Analysis based on CSO HFCS data. The typical gap for those landlords with insufficient funds to cover the €25k investment expenditure is approximately €14k or nearly half of the investment costs. This rises to €32k for the €50k investment, and €50k for the €75k investment using both measures. The typical gap for those landlords with insufficient funds to cover the €100k investment is €73k. 5.2.1 Cost of financing The final element in this section explores the cost implications for landlords if they were to borrow the financing for the investment gap. To do this, we undertake the following simulation. For each landlord who has a financing gap as defined above, we simulate the cost of the overall investment gap, in terms of the monthly repayments if the loan were to be financed using a personal loan. We use an 66|Investment barriers and the landlord structure However, these indicators are unlikely to capture the full extent of landlords with investment requirements in their own household. FIGURE 5.11 PERCENTAGE OF HOUSEHOLDS INDICATING A PROBLEM IN THEIR DWELLING Property is dark/lacks light Property has a problem with damp, leaking roof or rot Source: Analysis based on EU SILC data. To gain further insights into the risk appetite for investment among landlords, we can draw on the HFCS data. An indicator in the survey comes in the question that asks households to identify their willingness to take risks to earn rewards. This indicator is useful in that it may provide a proxy for landlords’ willingness to engage in energy efficiency expenditures with an uncertain return. Households are asked to indicate whether they are willing to take above-average risk to earn aboveaverage return, average risk to earn average return or no risk at all. This is a good indicator of investment appetite. We present the data below in Figure 5.12 for those landlords who are identified as having an investment gap in the €25k scenario. Approximately 41 per cent indicate they would be unwilling to take any risk, a further 37 per cent would take average risk and only 29 per cent would take above-average risk. This suggests quite a risk averse group of households. 5 3 6 9 0 5 10 15 Non-landlord Landlord Per cent of households Ireland Other countries 13 10 12 9 0 5 10 15 Non-landlord Landlord Per cent of households Ireland Other countries Investment requirements for energy efficiency upgrades in the rental sector|67 FIGURE 5.12 PROPORTION OF LANDLORDS WITH AN INVESTMENT GAP BY THEIR WILLINGNESS TO INVEST Source: Analysis based on CSO HFCS data. Note: Based on a scenario with an investment requirement of €25k. While we have raised the issue of landlord age as a potential discouraging factor in terms of investment propensity, a number of general broader factors may counteract this. From a purely financial perspective, any investment should be evaluated on a value for money basis, depending on the specific internal rate of return on the investment using discounted cash flows. This would incorporate an assessment of the investment costs, the cost of financing if required, the regulatory environment and the future rental price. As rents can be reset under the Rent Pressure Zone legislation once a major change in energy efficiency is secured, any landlord is likely to be able to re-price to recover the cost of the investment in financial terms. They can also avail of grant supports. These factors would likely offset some of the hesitancy in terms of risk appetite noted here. The impact of the perceived and actual level of compliance monitoring would also factor into their considerations. The complex interplay of all these factors is likely to determine the actual observed investment levels that will occur. 29.7% 37.2% 41.1% Above average risk Average risk No risk 68| Concluding remarks CHAPTER 6 Concluding remarks Retrofitting the Irish private rental sector (PRS) to reduce carbon emissions will be a critical component of the overall climate transition strategy for the residential housing sector. However, there are a number of notable challenges with the rental sector that raise the complexity of meeting these targets relative to the retrofit activity of homeowners. In particular, the existence of split incentives between landlord and tenant, as well as issues regarding the capacity of the landlord sector to finance and deploy the capital investments required, are distinct challenges, which differ from those presenting in other residential market cohorts. One starting point to measuring the scale of this challenge can be found in empirical estimates of the investment costs required to upgrade the sector, based on detailed information on the current energy efficiency status of rental properties. In this report, we aim to begin this process by achieving three main objectives. Firstly, we provide a profile of the sector in terms of its measured energy efficiency (across building energy rating (BER) categories). Secondly, we provide cost estimates (both at a property level and an aggregate level) as to what the investment cost would be to upgrade the housing stock concerned to a higher energy efficiency level. Thirdly, we explore the possible barriers to investment activity that may exist due to the high share of ‘household landlords’, and their potentially limited financial capacity to invest in energy efficiency upgrades. In doing so, we draw on a number of micro-level datasets. We used the Residential Tenancies Board (RTB) register of annual tenancies and the Sustainable Energy Authority of Ireland (SEAI) BER ratings database to proxy the stock of properties in the rental sector. Alongside this, we used Local Authority Social Housing Upgrade (LASHU) data and the SEAI One Stop Shop (OSS) information to cost energy efficiency upgrades. A final source was the Household Finance and Consumption Survey (HFCS). 6.1 FINDINGS ON INVESTMENT NEEDS Using a series of datasets (the RTB data for 2022, and SEAI data for various pooled years), we estimate that approximately 80 to 85 per cent of private rented dwellings currently have a BER rating below B; this constitutes approximately 240,000 to 260,000 properties. The vast majority of properties in this group have a D or C rating. In terms of the individual cost of upgrading properties to a B average rating, the data indicate an average cost of €43k for G-rated properties, and just under €30k per property for those currently C rated. Our estimated aggregate cost for the required upgrades in the sector ranges between €7bn and €8bn in total. Investment requirements for energy efficiency upgrades in the rental sector|69 These figures suggest a substantial investment requirement in order for the sector to meet the proposed B2 ratings for properties that are rented in the private market. The achievement of this aim will be dependent on a multitude of factors, including the structure of the landlord sector (individuals versus corporations), the availability of policy supports, the regulatory environment and landlord characteristics (e.g. lifecycle stage, access to finance, investment appetite). For example, corporate landlords are more likely to be able to use economies of scale and have sufficient financial resources to upgrade properties relative to household landlords. The investment decision will however be assessed in the context of a changing sector, with uncertain return profiles given the existence of rent controls. 17 The estimates provided in this research report contribute to the emergence of a clearer picture regarding the scale of this challenge. A number of caveats and limitations to our research must be kept in mind. First, we are working with mainly self-reported BER ratings; biases may exist in which case the actual BER distribution may differ from that outlined here. We do use the SEAI dataset as a secondary source for checking the robustness of the findings; nonetheless we cannot rule out this possibility entirely. Second, the sample of cost upgrades we use is relatively limited in terms of the number of observations and the focus on local authority activities. Again, this may lead to biases in our cost estimates. However, no other data source is available for the purposes of this study. Future research that addresses these data gaps would be welcome. A further issue relates to our methods used to match the cost estimates with the recorded change in BER ratings. As BER ratings are very dependent on the specific properties concerned, the development of a more detailed matching system, at the property level, rather than the use of a parsimonious method with a small number of characteristics would mark a future improvement. Finally, the deployment of additional scenarios (such as to an A-rated level) would be worthy of consideration, though they would be likely to raise the cost. 6.2 FINDINGS ON THE LANDLORD STRUCTURE AND FINANCIAL CAPACITY TO INVEST The findings of this research present a considerable challenge to the achievement of decarbonisation through investments in retrofit for the household landlord sector. International research has indicated that sub-optimal investments in energy efficiency in the rental sector occur quite frequently due to market failures and the split incentives in cost and yield on investment. 17 It is important to note that exemptions are in place for substantial renovations, including energy efficiency upgrades, although very few have been recorded (Coffey et al., 2022). 70| Concluding remarks There are reasons to believe that these factors are likely to be present for many Irish household landlords, with considerable barriers presenting to their investing in retrofit. Many landlords do not have the financial resources to make the simulated investments we deploy in this analysis when relying on their existing financial assets. Furthermore, there are likely to be both demand and supply side factors that act to inhibit investment in retrofit in such properties. On the demand side, some landlords may face financial challenges in their general personal circumstances: one in eight landlords has less than €50k total annual income. These landlords are unlikely to be in a position to commit cash flows towards investment, especially if the return on that investment is not borne by them. Secondly, many of these households are likely to have to also consider the retrofit of their main residential dwelling, which is another call on their resources in an energy efficiency context. From the supply side, access to credit in repayment terms could be challenging for these households if commercial finance is needed to cover the financing gaps, particularly given the older age profile of these landlords, a factor that reduces any potential payback period. Rising interest rates by global central banks, as part of the snapback in monetary policy due to high inflation, is also likely to raise the cost of financing investment in energy efficiency, to present viability challenges, and to further enhance split incentive issues. Low cost loans, such as the SBCI-backed Home Energy Loan Scheme, can be impactful in terms of lowering the cost of financing investments that come under its remit. This report focuses on filling a knowledge gap in terms of the current energyefficiency profile of the PRS, likely upgrade costs and household landlords’ financial position. Its findings suggest that policymakers face a very complex challenge in relation to encouraging these landlords to engage in energy efficiency upgrades on their rented properties. It is beyond the study’s scope to assess the numerous policies, grants and supports that are currently available, and their potential role in addressing such financing gaps, as well as any issues regarding the willingness of landlords to address the challenges identified. These are crucial topics for future research. The focus of this report has been on the current state of play in terms of energy efficiency and landlords’ financial situation, and incentives. However, the major implications of energy upgrade requirements for tenants, both in terms of likely monetary costs and tenancy security, must be kept in mind. Indeed, some of the most vulnerable tenants in the PRS live in the least energy efficient properties at present; Housing Assistance Payment (HAP) and Rent Supplement tenants have the highest share of properties with a BER rating of F or G. It is important that any policy interventions take a balanced approach to rental sector upgrades. This includes weighing up the need to make progress, but also the dangers this process might pose for tenants in particular, as it could potentially result in a reduction of Investment requirements for energy efficiency upgrades in the rental sector|71 the rental stock at a time where there are already significant shortages relative to demand. 72|References REFERENCES Allcott, H. and M. Greenstone (2012). ‘Is there an energy efficiency gap?’, Journal of Economic Perspectives, Vol. 26, No. 1, pp. 3–28, http://dx.doi.org/10.1016/B9780-12-397879-0.00005-0. Ambrose, A.R. (2015). ‘Improving energy efficiency in private rented housing: Why don’t landlords act?’, Indoor and Built Environment, Vol. 24, No. 7, pp. 913–924, https://doi.org/10.1177/1420326X15598821. Ástmarsson, B., P.A. Jensen and E. Maslesa (2013). ‘Sustainable renovation of residential buildings and the landlord/tenant dilemma’, Energy Policy, Vol. 63, pp. 355–362, https://doi.org/10.1016/j.enpol.2013.08.046. Byrne, M. and R. McArdle (2022). ‘Secure occupancy, power and the landlord–tenant relation: A qualitative exploration of the Irish private rental sector’, Housing Studies, Vol. 37, No. 1, pp. 124–142, https://doi.org/10.1080/02673037.2020.1803801. Carroll, J., C. Aravena and E. Denny (2016). ‘Low energy efficiency in rental properties: Asymmetric information or low willingness-to-pay?’, Energy Policy, Vol. 96, pp. 617–629, https://doi.org/10.1016/j.enpol.2016.06.019. Castellazzi, L., P. Bertoldi and M. Economidou (2017). Overcoming the split incentive barrier in the building sector, Publications Office of the European Union, Luxembourg, 10, 912494. Charlier, D. (2015). ‘Energy efficiency investments in the context of split incentives among French households’, Energy Policy, Vol. 87, pp. 465–479, https://doi.org/10.1016/j.enpol.2015.09.005. Coffey, C., P.J. Hogan, C. O’Toole, K. McQuinn and R. Slaymaker (2022). Rental inflation and stabilisation policies: International evidence and the Irish experience, ESRI Research Series Report No. 136, Dublin: ESRI, https://doi.org/10.26504/rs136. Collins, M. and J. Curtis (2018a). ‘Rental tenants’ willingness-to-pay for improved energy efficiency and payback periods for landlords’, Energy Efficiency, Vol. 11, No. 8, pp. 2033–2056, https://doi.org/10.1007/s12053-018-9668-y. Collins, M. and J. Curtis (2018b). ‘Willingness-to-pay and free-riding in a national energy efficiency retrofit grant scheme’, Energy Policy, Vol. 118, pp. 211–220, https://doi.org/10.1016/j.enpol.2018.03.057. Cornago, E. and L. Dressler (2020). ‘Incentives to (not) disclose energy performance information in the housing market’, Resource and Energy Economics, Vol. 61, https://doi.org/10.1016/j.reseneeco.2020.101162. Corrigan, E., D. Foley, K. McQuinn, C. O’Toole and R. Slaymaker (2019). ‘Exploring affordability in the Irish housing market’, The Economic and Social Review, Vol. 50, Issue 1, pp. 119–157. Coyne, B. (2023). Residential retrofit review, Climate Change Advisory Council Working Paper Series No. 18, Climate Change Advisory Council. Investment requirements for energy efficiency upgrades in the rental sector|73 Daly, P. (2023). ‘Institutional investment in housing: Financialisation 2.0 in the case of Ireland’, Journal of the Statistical and Social Inquiry Society of Ireland, Vol. 52, 2022/23, pp. 60–82. Fuerst, F., M. Haddad and H. Adan (2020). ‘Is there an economic case for energy-efficient dwellings in the UK private rental market?’, Journal of Cleaner Production, Vol. 245, 118642, https://doi.org/10.1016/j.jclepro.2019.118642. Gerardi, K., K.F. Herkenhoff, L.E. Ohanian and P.S. Willen (2018). ‘Can’t pay or won’t pay? Unemployment, negative equity, and strategic default’, The Review of Financial Studies, Vol. 31, No. 3, pp. 1098–1131, https://doi.org/10.1093/rfs/hhx115. Government of Ireland (2021). Climate Action Plan 2021: Securing Our Future, Dublin: Department of the Environment, Climate and Communications. Government of Ireland (2024). Climate Action Plan 2024, Dublin: Department of the Environment, Climate and Communications. Hope, A.J. and A. Booth (2014). ‘Attitudes and behaviours of private sector landlords towards the energy efficiency of tenanted homes’, Energy Policy, Vol. 75, Issue C, pp. 369–378. Jaffe, A.B. and R.N. Stavins (1994). ‘The energy-efficiency gap – What does it mean?’, Energy Policy, Vol. 22, No. 10, pp. 804–810, https://doi.org/10.1016/03014215(94)90138-4. Kristopher, G., K.F. Herkenhoff, L.E. Ohanian and P.S. Willen (2018). ‘Can’t pay or won’t pay? Unemployment, negative equity, and strategic default’, The Review of Financial Studies, Vol. 31, No. 3, pp. 1098–1131, https://doi.org/10.1093/rfs/hhx115. Lang, M., R. Lane, K. Zhao, S. Tham, K. Woolfe and R. Raven (2021). ‘Systematic review: Landlords willingness to retrofit energy efficiency improvements’, Journal of Cleaner Production, Vol. 303, https://doi.org/10.1016/j.jclepro.2021.127041. Miu, L. and A.D. Hawkes (2020). ‘Private landlords and energy efficiency: Evidence for policymakers from a large-scale study in the United Kingdom’, Energy Policy, Vol. 142, Issue C, https://doi.org/10.1016/j.enpol.2020.111446. Nie, H., R. Kemp, J. Xu, V. Vasseur and Y. Fan (2020). ‘Split incentive effects on the adoption of technical and behavioral energy-saving measures in the household sector in Western Europe’, Energy Policy, Vol. 140, Issue C, https://doi.org/10.1016/j.enpol.2020.111424. Maruejols, L. and D. Young (2011). ‘Split incentives and energy efficiency in Canadian multifamily dwellings’, Energy Policy, Vol. 39, No. 6, pp. 3655–3668, https://doi.org/10.1016/j.enpol. McQuinn, K., C. O’Toole and R. Slaymaker (2021). ‘Credit access, macroprudential rules and policy interventions: Lessons for potential first time buyers’, Journal of Policy Modeling, Vol. 43, No. 5, p. 944–963. Murphy, L., F. Meijer and H. Visscher (2012). ‘A qualitative evaluation of policy instruments used to improve energy performance of existing private dwellings in the Netherlands’, Energy Policy, Vol. 45, pp. 459–468, https://doi.org/10.1016/j.enpol.2012.02.056. 74|References Myers, Erica (2020). ‘Asymmetric information in residential rental markets: Implications for the energy efficiency gap’, Journal of Public Economics, Vol. 190, 104251, https://doi.org/10.1016/j.jpubeco.2020.104251. Petrov, I. and L. Ryan (2021). ‘The landlord-tenant problem and energy efficiency in the residential rental market’, Energy Policy, Vol. 157, https://doi.org/10.1016/j.enpol.2021.112458. Pillai, A., M.T. Reaños and J. Curtis (2021). ‘An examination of energy efficiency retrofit scheme applications by low-income households in Ireland’, Heliyon, Vol. 7, No. 10, https://doi.org/10.1016/j.heliyon.2021.e08205. Residential Tenancies Board (2023). RTB rental sector survey – Small landlords report, Dublin: Residential Tenancies Board. Slaymaker, R., B. Roantree, A. Nolan and C. O’Toole (2022). Future trends in housing tenure and the adequacy of retirement income, ESRI Research Series Report No. 143, Dublin: ESRI, https://doi.org/10.26504/rs143. Slaymaker, R. and E. Shiel (2023). New vs existing rental tenancies in Ireland: a first look at annual registrations microdata, ESRI Survey and Statistical Report Series 122, Dublin: ESRI, https://doi.org/10.26504/sustat122. Society of St. Vincent de Paul Ireland (2015). Energy efficiency of rental accommodation in Ireland, Tech. rep. McDowell and Purcell Solicitors, pp. 1–27,. Society of St. Vincent De Paul and Threshold (2021). Warm housing for all: Strategies for improving energy efficiency in the private rented sector. Weber, I. and A. Wolf (2018). ‘Energy efficiency retrofits in the residential sector: analysing tenants’ cost burden in a German field study’, Energy Policy, Vol. 122, pp. 680–688, https://doi.org/10.1016/j.enpol.2018.08.007. Zeitler, J.-A. (2018). ‘H2020 – RentalCal – European rental housing framework for the profitability calculation of energy efficiency retrofitting investments’, Journal of Property Investment & Finance, Vol. 36, No. 1, pp. 125–131, https://doi.org/10.1108/09574090910954864. Appendix A|75 APPENDIX A Additional results TABLE A.1 CONCORDANCE OF DWELLING TYPES Census 2022 RTB dataset SEAI BER register SEAI OSS LASHU data Detached house Whole house, detached part house, detached Detached house House Detached house Other Semidetached house Whole house, detached part house, detached Semi-detached house Semi-detached/End terrace Terraced house Whole house, terraced part house, terraced Mid-terrace house End of terrace house Mid terrace Flat or apartment, purpose-built flat or apartment, converted Bed-sit Apartment Flat Maisonette, semi-det Maisonette, terraced Maisonette, detached Bed-sit Apartment Maisonette Top-floor apartment Mid-floor apartment Ground-floor apt. Basement dwelling Apartment Apts Notes: In LASHU classification, ‘apts’ includes apartments and mid-terrace dwellings. TABLE A.2 MEAN ESTIMATES OF THE RENTAL HOUSING STOCK SIZE Estimates with accounting for BER-exempt dwellings Estimates without accounting for BER-exempt dwellings RTB SEAI RTB SEAI BER Est. num. % Est. num. % Est. num. % Est. num. % G 9,663 2.9 15,687 4.7 10,229 3.1 16,693 5 F 9,523 2.9 14,087 4.3 10,048 3 14,853 4.5 E2 10,239 3.1 14,897 4.5 10,808 3.3 15,732 4.8 E1 16,005 4.8 19,138 5.8 16,995 5.1 20,283 6.1 D2 31,136 9.4 36,874 11.2 32,935 10 38,930 11.8 D1 40,708 12.3 42,095 12.7 42,848 13 44,304 13.4 C3 45,029 13.6 43,022 13 47,364 14.3 45,224 13.7 C2 38,859 11.8 42,254 12.8 40,923 12.4 44,554 13.5 C1 41,306 12.5 37,019 11.2 43,725 13.2 39,277 11.9 B3 24,767 7.5 25,326 7.7 26,394 8 27,055 8.2 B2 13,054 3.9 11,678 3.5 14,068 4.3 12,573 3.8 B1 8,324 2.5 3,696 1.1 8,938 2.7 3,996 1.2 A3 8,989 2.7 4,092 1.2 9,485 2.9 4,310 1.3 A2 13,014 3.9 2,637 0.8 13,832 4.2 2,809 0.8 A1 1,922 0.6 37 0 2,040 0.6 40 0 Exempt 18,095 5.5 18,095 5.5 - - - - Total 330,632 100 330,632 100 330,632 100 330,632 100 Subtotals F/G 19,186 6 29,774 9 20,277 6 31,546 10 E* 26,243 8 34,035 10 27,804 8 36,015 11 D* 71,844 22 78,969 24 75,783 23 83,234 25 C* 125,195 38 122,295 37 132,011 40 129,055 39 B* 46,145 14 40,699 12 49,399 15 43,624 13 A* 23,925 7 6,765 2 25,358 8 7,159 2 Exempt 18,095 5 18,095 5 - - - - Total 330,632 100 330,632 100 330,632 100 330,632 100